Setting up the SIGNLD extension for Claude: Explained

Setting up the SIGNLD extension for Claude has two parts: an admin connects SIGNLD's Knowledge Graph and enables the extension once, and each user then adds it to their own Claude session. Both sides stay read-only, and every fact the extension surfaces stays traceable to a source record.

This guide separates the two roles deliberately because the setup work is genuinely different: an admin makes a workspace-level decision about scope and access, while a user makes a one-time personal decision to authenticate their own Claude session. Confusing the two steps is the most common reason setup stalls.

For the wider context, see our overview of the SIGNLD extension for Claude.

Prerequisites

Before starting, confirm three things. First, a SIGNLD account with at least one connected system and an established Knowledge Graph, since the extension has nothing to expose to Claude until SIGNLD has mapped some business context. Second, a Claude account with extension support enabled for the workspace: the exact plan and workspace requirements for extensions live on Anthropic's documentation, because they change and SIGNLD does not control them. Third, an admin role in SIGNLD, since enabling the extension for a workspace is an admin-level action, not something an individual user can turn on for the whole team.

The SIGNLD extension for Claude is built and maintained by SIGNLD using Anthropic's publicly available extension interfaces. SIGNLD is not affiliated with, sponsored by, or endorsed by Anthropic.

What the admin does

An admin sets this up once, for the whole workspace:

Sign in to SIGNLD and open the extension settings from the admin panel.
Enable the SIGNLD extension for Claude for the workspace, which generates the connection credentials the extension needs to authenticate with SIGNLD.
Choose which connected systems and which parts of the Knowledge Graph are in scope for the extension. An admin can scope this narrower than the full graph if only certain teams should have Claude access to certain data.
Set which users or roles are allowed to use the extension, following the same permission model SIGNLD already applies to the rest of the platform.
Confirm the connection is read-only, which it is by default and cannot be changed to write access from the extension.

Once this is done, the admin's job is mostly maintenance: reviewing who has access, updating scope as new systems get connected to SIGNLD, and removing access when someone leaves a team.

What the user does

Each individual user then connects their own Claude session, following Anthropic's own steps for adding an extension in Claude, which are documented at docs.anthropic.com since the exact menus and screens in Claude are Anthropic's to describe and can change. In outline, a user finds the SIGNLD extension in Claude's extension list, adds it to their session, and authenticates with their SIGNLD account so Claude can request context that SIGNLD is allowed to share with that user specifically.

After that one-time authentication, the user does nothing extra per session. They ask a question in Claude the way they normally would, and if the answer needs business context SIGNLD holds, the extension supplies it in the background.

How it works, step by step

The admin enables the extension and sets scope, as above.
A user connects the extension to their Claude session and authenticates.
During a Claude session, when a question touches business context, Claude can call the SIGNLD extension to look up relevant entities, metrics, rules, or definitions from the Knowledge Graph.
SIGNLD returns only what that user is permitted to see, based on the same access rules that apply everywhere else in SIGNLD.
Claude uses that context to shape its answer, and the underlying facts stay traceable back to the source record in SIGNLD.

Nothing in this flow changes a source system. The extension reads from the Knowledge Graph and returns information; it does not write, update, or trigger anything downstream.

What changes for the person asking the question

Before the extension is set up, a user asking Claude a business question has to supply the context by hand: pasting numbers, describing definitions, explaining which system a figure came from. After setup, that context can come from SIGNLD directly, so the person still asks the question in their own words, but spends less time assembling the background Claude needs and gets an answer grounded in the same entities and records SIGNLD already tracks.

Permissions and data access

The extension inherits SIGNLD's existing permission model. A user only sees what they're already allowed to see in SIGNLD; the extension does not grant broader access than the platform already grants that person. Access is read-only against every connected source system, and nothing the extension surfaces is disconnected from where it came from: each piece of context can be traced back to the record it originated from in SIGNLD. Admins can review and revoke extension access for any user at any time from the same panel used to manage other SIGNLD permissions.

Troubleshooting

The extension does not appear in Claude's extension list. Confirm the admin has enabled the extension for the workspace and that the signed-in user's workspace matches the one where it was enabled. If it still doesn't appear, check Anthropic's documentation for workspace-level extension requirements, since those requirements are set by Anthropic and can affect visibility.

Authentication fails when connecting the extension. Confirm the user has an active SIGNLD account and has been granted access by an admin. A user who isn't on the allowed list for the extension will not be able to complete authentication even if the extension itself is visible.

Claude gives an answer without any SIGNLD context. This usually means the question didn't touch anything mapped in the Knowledge Graph, or the relevant system hasn't been connected to SIGNLD yet. Check which systems are connected and in scope for the extension, and confirm the entity or metric in question actually exists in the graph.

Context looks out of date. SIGNLD's Knowledge Graph reflects the state of connected systems as of its last sync. If a source system changed recently, confirm the connection is syncing normally in SIGNLD before assuming the extension is at fault.

Access was revoked but the user can still see old answers. Revoking access stops new lookups; it does not retroactively erase answers already generated in a past Claude session, since SIGNLD does not control Claude's session history.

Setup is a one-time cost for both the admin and each user. After the first authentication, no one needs to repeat these steps unless access is revoked or the extension is disabled and re-enabled for the workspace.

Key takeaways

An admin sets this up once, for the whole workspace:.
Nothing in this flow changes a source system.
Setup is a one-time cost for both the admin and each user.
Plan and workspace requirements for using extensions in Claude are set by Anthropic and documented at docs.anthropic.com.
Issues with Claude's interface, authentication screens, or extension list are Anthropic's to support; their documentation and support channels are at docs.anthropic.com.

FAQ

Do I need a specific Claude plan to use the SIGNLD extension?

Plan and workspace requirements for using extensions in Claude are set by Anthropic and documented at docs.anthropic.com. SIGNLD does not publish or guarantee Claude plan details because they are outside SIGNLD's control and can change.

Can a user enable the extension for the whole team themselves?

No. Enabling the extension for a workspace is an admin-level action in SIGNLD. Individual users connect their own Claude session to an extension the admin has already turned on and scoped.

Does the extension let Claude change anything in our connected systems?

No. The connection is read-only in both directions: SIGNLD reads from connected source systems, and the extension only returns information to Claude. It does not write back to SIGNLD or to any source system.

What happens if an admin narrows the scope after users are already connected?

Users immediately see the narrower scope on their next lookup. SIGNLD applies scope and permission checks at the time context is requested, not only at setup.

Is setup different for each connected source system?

No. Setup is done once against SIGNLD's Knowledge Graph, which already unifies the connected systems. There's no separate extension setup per source system.

Who do I contact if something in Claude's own interface isn't working?

Issues with Claude's interface, authentication screens, or extension list are Anthropic's to support; their documentation and support channels are at docs.anthropic.com. Issues with what the extension returns or with SIGNLD access are SIGNLD's to support.

Learn more about the SIGNLD extension for Claude, how SIGNLD's Knowledge Graph works, and read about security.

Try SIGNLD free or see how it works.

Claude and Anthropic are trademarks of Anthropic. Used descriptively only.

Setting up the SIGNLD extension for Claude

In this article